[ZEPPELIN-1894] Matplotlib is very slow in python interpreter

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.7.0
• Fix Version/s: 0.7.1, 0.8.0
• Assignee: Shim HyungSung
• Reporter: Jeff Zhang
• Created: 03/Jan/17 12:55
• Updated: 18/Mar/17 16:35
• Resolved: 18/Mar/17 16:35

Description

The following code takes 20 seconds in zeppelin python interpreter. But it only take 1 second in jupyter.

import matplotlib.pyplot as plt
import numpy as np

x = np.random.normal(0,1,1024)
y = np.random.normal(0,1,1024)

plt.scatter(x,y, s=25)
plt.xlim((-2,2))
plt.ylim((-2,2))

It looks like most of the time is spent on sending image to client.

DEBUG [2017-01-03 21:02:05,284] ({pool-2-thread-2} PythonInterpreter.java[sendCommandToPython]:248) - Sending :
import matplotlib.pyplot as plt
import numpy as np

x = np.random.normal(0,1,1024)
y = np.random.normal(0,1,1024)

plt.scatter(x,y, s=25)
plt.xlim((-2,2))
plt.ylim((-2,2))
z._displayhook()
DEBUG [2017-01-03 21:02:05,507] ({pool-2-thread-2} PythonProcess.java[sendAndGetResult]:114) - Read line from python shell : <matplotlib.collections.PathCollection object at 0x106f5eb10>
DEBUG [2017-01-03 21:02:05,532] ({pool-2-thread-2} PythonProcess.java[sendAndGetResult]:114) - Read line from python shell : (-2, 2)
DEBUG [2017-01-03 21:02:05,557] ({pool-2-thread-2} PythonProcess.java[sendAndGetResult]:114) - Read line from python shell : (-2, 2)
DEBUG [2017-01-03 21:02:05,557] ({pool-2-thread-2} PythonProcess.java[sendAndGetResult]:114) - Read line from python shell : %html
DEBUG [2017-01-03 21:02:05,619] ({pool-2-thread-2} PythonProcess.java[sendAndGetResult]:114) - Read line from python shell : <div style='width:auto;height:auto'><img src=data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAlgAAAGQCAYAAAByNR6YAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAIABJREFUeJzsvX10XFd57/8djzU+9agXTAHHmLx45ER2pNFoHGJhglcTGWQUjxteTImjAdZiFegqK21viiXnxmNJIy4Jt782LeUPktWWdaGshEscTe4omcrtunF/rFxbWU1kYi7ppUkI...
DEBUG [2017-01-03 21:02:05,620] ({pool-2-thread-2} PythonInterpreter.java[sendCommandToPython]:254) - Got :
<matplotlib.collections.PathCollection object at 0x106f5eb10>
(-2, 2)
(-2, 2)
%html..

INFO [2017-01-03 21:02:19,754] ({pool-2-thread-2} SchedulerFactory.java[jobFinished]:137) - Job remoteInterpretJob_1483448524770 finished by scheduler org.apache.zeppelin.python.PythonInterpreter789681602

Issue Links

• duplicates: ZEPPELIN-1360 Printing long strings is very slow in the python interpreter
• is duplicated by: ZEPPELIN-2160 PySpark: Matplotlib Integration extremely slow
• links to: GitHub Pull Request #1931